by Chris Byrne (written 31.7.25)
Subscribe to my free newsletter: https://seoetc.beehiiv.com/subscribe ( I might share the prompts and actual full responses)...
The above video is a discussion created using Google 's Notebook LM relating to my attempt to reverse engineer a response to a comparative entity prompt ("best smartphones") in both Chatgpt and Google Gemini with a simple overarching prompt . The video looks at the intitial prompt (to generate the 'normal' response), comparing and contrasting the 'normal' responses and the outline of the process used to generate the individual original responses in Chatgpt and Gemini in detail. I've edited some process names out of the discussion.
In the USA for example, there are 10s of thousand of "best smartphone"-type searches on Google alone, see a sample below:
I cannot be certain the responses do not contain hallucination, but for Chatgpt (the current focus of my research) the result are consistent with my research (on LLMs) hitherto.
The responses outlined the processes used to generate responses for a comparative entity-type prompt.
Subscribe to my free newsletter: https://seoetc.beehiiv.com/subscribe .
See my related article: https://www.seolondonsurrey.co.uk/blog/mapping-entity-associations-in-chatgpt.